Automatic Identification System of Silkworm Cocoon Based on Computer Vision Method

With the continuous improvement of image recognition technology, people began to transfer their research to the sex recognition of silkworm cocoons. However, due to the inaccurate identification method and low efficiency, it has been difficult to develop a perfect identification system. In order to accurately and efficiently perform automatic gender recognition on silkworm cocoons, this paper proposes a multi-resolution local Gabor binary pattern (MLGBP) feature extraction method based on computer vision to comprehensively describe the fine and rough local microscopic patterns of silkworm cocoons. The experimental results show that, in the vast majority of cases, MLGBP achieved an accuracy of at least 95% and a maximum classification accuracy of 98.8%. This paper proposes a gender classification method based on feature information based on computer vision technology. The experimental results on the AR, CAS-PEAL and FERET databases show that SVM can improve the classification accuracy compared to the single feature. In this paper, the classifier is designed by artificial neural network theory, and BP neural network is used for classification and recognition. In the experiment, 10 samples of silkworm pupa of 871A variety were selected as the training set, and the classification training was carried out. The experimental results showed that the recognition rate reached 90%.